Optical Signal Processing with Reduced Power Consumption

Due to an exponential increase in the number of consumers using the voice, video and data services associated with modern optical transport systems, the footprint of these systems can no longer be ignored. To that end, this thesis focuses on studying the nonlinear effects in highly nonlinear fibers (HNLFs) and semiconductor optical amplifiers (SOAs) along with thermally induced power saturation effects in vertical-cavity surface-emitting lasers (VCSELs) from the standpoint of reducing power consumption associated with wavelength converter and optical signal regenerator topologies based on these devices. This thesis begins with a numerical investigation into the interplay between dispersion and nonlinearity for optimizing the performance and reducing the input power requirement of an all-optical re-amplification and re-shaping (2R) regenerator based on self-phase modulation (SPM) and spectral filtering at 40 Gb/s. Careful dispersion optimization leads to 30% reduction in input power required for HNLF-based 2R-regenerator without significant performance degradation. However, there are some associated trade-offs and watt level input optical powers are still required.;This thesis therefore shifts focus towards amplified spontaneous emission (ASE)-induced enhancement in nonlinear effects [SPM and four-wave mixing (FWM)] in SOAs. Traditionally, SOAs have required few tens of milliwatts (mWs) of input powers for optical signal processing. The modern SOAs investigated in this thesis achieve high internal ASE at high bias currents which reduces the effective carrier lifetime down to ≈ 10 ps. A simplified numerical model is developed for these devices which is used to study SPM and FWM. For the case of SPM, a trade-off between spectral broadening and spectral symmetry due to ASE-mediated gain dynamics is demonstrated. This thesis also investigates the conversion efficiency (CE) and optical signal to noise ratio (OSNR) properties of ASE-assisted FWM in SOAs and demonstrates that in the presence of high internal ASE, the requirement on input optical powers for wavelength conversion is reduced from 10s of mWs to less than 1 mW. A 10-Gb/s L-band wavelength converter based on FWM is also realized experimentally.;While ASE-enhanced nonlinear effects in SOAs do relax the requirements on input signal powers, the requirement of high bias currents places a serious question mark on the scalability of this technique. VCSELs or vertical cavity SOAs (VCSOAs) (low-Q version of standard VCSELs) consume a fraction of electrical power as compared to SOAs. However the high speed operation of these devices is limited to a great extent by the thermally-induce power saturation effects. This thesis develops an empirical thermal model which can be used for studying mechanisms contributing to thermally induced optical power rollover in VCSELs.;The model is applied to four oxide-confined, 850 nm VCSEL designs. For all these devices, it is demonstrated that the power dissipation due to linear power dissipation exceeds power dissipation across the series resistance at any ambient temperature and bias current. The dominant contributors to device heating for these devices are the power dissipated across series resistance and carrier leakage. Absorption heating is also quite significant in the absence of a shallow surface-etch in the top DBR. Carrier leakage places the ultimate limit on the thermal performance for this entire class of devices. Further, a trade-off between power dissipated due to optical absorption and carrier leakage in these devices is demonstrated and investigated. The empirical thermal model proposed in this thesis is universally applicable for identifying the mechanisms limiting the thermal performance and for formulating the design strategies to ameliorate them.